有人会问，即便如此，价格的变动也会通过偏好关系进而改变每个人的需求量吧。这样一来，产量不就变了？确实，但这并不先验地意味着边际主义的一般均衡模型就是正确的：
首先，古典可以通过“两步走”策略继续分开研究价格与产量决定，进而兼容价格与产量之间存在互动的事实：例如工资率发生变动，我们要研究这一变动的影响。那么，我们可以先假定产量不变，研究工资变动发生的立刻效果。然后在这个基础之上，再假定工资固定在新位置，研究此时产量如何因为工资率的变动（通过偏好关系）而变动以及这种变动带来的影响。
其次。一旦我们抛弃有缺陷的边际主义体系，那么这种两步走策略就是必需的。这也意味着价格与需求量之间没有明确的函数关系，不能像边际主义那样直接用偏好确定价格。偏好应当被视为确定相对价格的众多间接因素之一：
“More generally, once marginal theory is abandoned, it no longer seems possible to determine individual incomes, and therefore the quantities demanded, by means of functional relations having properties that are sufficiently definite and at the same time sufficiently general as to render the procedure meaningful at the level of a general theory, as distinct from special 'models' (see para. 22).52 It would therefore seem inevitable to proceed by separate logical stages, as indicated above.
更一般的说，一旦边际主义理论被抛弃，就不可能通过拥有充分明确且同时充分一般的函数关系去确定个体收入和需求量，从而使程序在有别于特殊模型的一般理论的层次上具有意义。因此使用上文所示的分离逻辑步骤看起来是必然的。”
附录：Garegnani 论古典体系如何处理产量与价格的互动
Garegnani 在 Classical VS Marginalist Analysis 这篇文章已经指出，对于古典的产量决定理论而言，重要的是偏好的内容（content）而不是偏好的形式性质（formal property）：
“30 Outputs, however, also depend on what is modernly described as consumer preferences. With respect to this fourth circumstance, it is essential to start by distinguishing between two aspects of consumer preferences, which are generally taken together. The first aspect is the content of these preferences- the fact, that is, that at some given relative prices one individual may consume, say, one of a commodity A and nine of commodity B, rather than nine of A and one of B. The second aspect is the formal property for which when the relative price of a commodity A falls, the quantity of it consumed would increase. In the terms of modern theory, the distinction might be seen to correspond to that between the 'position' of a demand curve and its 'slope'. Now what is principally relevant in determining output is the first aspect of consumers' tastes, their 'content'.
30 然而，产出也取决于现代所说的消费者偏好。关于第四种情况，首先必须区分消费者偏好的两个方面，这两个方面通常是结合在一起的。第一个方面是这些偏好的内容，也就是说，在某些给定的相对价格下，一个人可能会消费一个商品A和九个商品B，而不是九个商品A和一个商品B。第二个方面是形式性质，当商品a的相对价格下降时，它被消费的数量会增加。根据现代理论，这种区别可能被视为对应于需求曲线的“位置”及其“斜率”之间的区别。现在，决定产出的主要相关因素是消费者口味的第一个方面，即他们的“内容”。”
Garegnani 接着说，现代边际主义理论转而研究偏好的第二个方面而不是第一个（很明显，边际效用、边际替代率就是第二方面的概念！）：
“However, this is just what modern theorists leave to the analysis of psychologists or sociologists and take as given (under the guise of 'given tastes') when conducting their analysis of demand- which is exclusively concerned with the second aspect, the formal property.
然而，这正是现代理论家留给心理学家或社会学家分析的东西，他们在进行需求分析时（在“给定的品味”的幌子下）把它当作给定的东西，而需求分析只涉及第二个方面，即形式性质。”
Garegnani 接着告诫道，古典并不反对对偏好第一方面的分析。事实上，与边际主义相反，古典体系反而将对偏好的内容分析视为是重要的：
”It is therefore difficult to see why taking the 'content' of tastes as given should arouse objections when adopted in a classical context. The paradox, however, is that, unlike most modern theorists, classical authors considered the analysis of the 'content' of preferences to be an important part of economics (for example, in the analysis of the cultural elements affecting 'workers' necessaries', para. 12 above). This analysis was however conducted when dealing with wages and outputs and, to that extent, the outputs could be taken as given when determining prices.
因此，很难理解为什么在古典语境中采用品味的“内容”会引起反对。然而，矛盾之处在于，与大多数现代理论家不同，古典作家认为对偏好“内容”的分析是经济学的重要组成部分（例如，在分析影响“工人必需品”的文化因素时，上文第12段）。然而，这一分析是在处理工资和产出时进行的，在这种程度上，产出可以在确定价格时视为给定值。“
接下来，Garegnani 告诉我们可以通过两步走战术在古典框架下研究由相对价格变动引发的产量变动，以及由产量变动引发的相对价格变动：
“Where relevant, the dependence of outputs on prices can in fact be analysed in two successive stages: (i) The effect on prices, and on the dependent distributive variables, of the change in one or other of the independent variables (for example, in the real wage) is conducted while supposing outputs to be constant - or to undergo the assumed independent change when the outputs themselves are the changing independent variable (think of Ricardo's analysis of the extension of the cultivation of 'corn' to inferior lands). (ii) The effects on outputs of that change in relative prices can then be analysed in accordance with the circumstances of the case - together with the effects resulting directly from the initial change of the independent variable (for example, the changes in wage goods outputs resulting from the change in the real wage even if the relative prices of commodities were not to change). At this stage, it will also be possible to analyse the effects of the changed outputs on relative prices and on the dependent distributive variables, when non-constant returns to scale are assumed50 - together with any further effects that the changed prices might have on the independent variables. 51
在相关情况下，产出对价格的依赖性实际上可以分为两个连续阶段进行分析： (i) 首先，在假设产出保持不变的情况下，分析一个或多个独立变量（例如实际工资）变化对价格及其相关的分配变量的影响。当产出本身是变化的独立变量时，比如李嘉图分析耕种“谷物”扩展到较差土地的情形，就会假设产出会发生预设的独立变化。 (ii) 然后，根据具体情况，分析这种相对价格变化对产出的影响，以及由于最初独立变量的变化直接导致的影响（例如，即使商品的相对价格没有变化，实际工资的变化也会导致工资商品产出的变化）。在这个阶段，还可以分析产出变化对相对价格和依赖分配变量的影响，尤其是在假设规模报酬不恒定的情况下，以及变化的价格可能对独立变量产生的进一步影响。”
最后，虽然 Garegnani 承认两步走战术初看不如边际主义同时决定价格和产量的理论模型有吸引力，但是进一步的思考会发现这才是必需的：
“Any demand function has to depend on individual incomes. Now, in marginal theories the condition relating to the equilibrium in the market for factors (or other equivalent conditions, for example, in the so-called 'dis-equilibrium' theories) ensures that, as the demand price varies, the associated constancy or variation in individual incomes can be simultaneously determined and that the influence of this on the quantity demanded of the commodity can accordingly be reckoned. But, as we abandon marginal theory, we no longer have any such condition relating to factor markets and therefore, it seems, no sufficiently general ground for determining the change in the quantity demanded associated with a given change in price.
任何需求函数都必须依赖于个人收入。在边际理论中，与要素市场均衡相关的条件（或其他等效条件，例如所谓的“非均衡”理论中的条件）确保，随着需求价格的变化，个人收入的不变或变化可以同时被确定，因此可以相应地估算出这对所需商品数量的影响。但是，一旦我们放弃边际理论，我们就不再有任何与要素市场相关的条件，因此似乎也没有足够普遍的基础来确定价格变化所关联的需求量变化。”
“More generally, once marginal theory is abandoned, it no longer seems possible to determine individual incomes, and therefore the quantities demanded, by means of functional relations having properties that are sufficiently definite and at the same time sufficiently general as to render the procedure meaningful at the level of a general theory, as distinct from special 'models' (see para. 22).52 It would therefore seem inevitable to proceed by separate logical stages, as indicated above.
更广泛地说，一旦放弃边际理论，似乎就不再可能通过具有足够明确和同时足够普遍的函数关系来确定个人收入，因此也无法确定需求量，使得这种做法在一般理论层面上具有意义，而不仅仅是特殊“模型”层面上（参见第22段）。因此，似乎不可避免地需要按照上述的分别逻辑阶段进行。”
“This conclusion is strengthened by a second consideration. The received general concept of a dependence of the quantity demanded of a commodity on its price may be seen to run the risk of falling between two stools. If the effect of the price on the quantity bought is not appreciable, then the effect can be ignored without great error. Alternatively, when the effect is important enough to need general consideration, it seems it will often be the case that the effect constitutes an irreversible change, which is incompatible with its treatment in terms of a demand function. That is, the effect will entail a permanent change in the habits of consumers, which even marginalist authors would have to treat as a change of 'tastes' and therefore by a stage of analysis separate from the 'general equilibrium' determination of prices and the associated demand functions, where tastes appear as given. (Think, for instance, of the increase in the demand for cars in the USA in the 1920s, as technical developments led to an appreciable fall in their price.)
这一结论还因第二个考虑而得到加强。商品需求量依赖于其价格的一般概念可能会陷入两难境地。如果价格对购买数量的影响不显著，那么忽略这种影响并不会造成严重错误。另一方面，当这种影响足够重要以至于需要普遍考虑时，似乎通常情况下这种影响会构成一种不可逆转的变化，这与用需求函数来处理它不相容。也就是说，这种影响将导致消费者习惯的永久性变化，即使是边际主义作者也不得不将其视为“口味”的变化，并因此在“一般均衡”价格决定和相关需求函数的分析之外进行单独阶段的分析，其中口味被视为既定的。（例如，想想20世纪20年代美国汽车需求的增加，随着技术的发展，汽车价格显著下降。）”
